Method for controlling a vehicle

Abstract

The invention relates to a method for controlling a vehicle (1) along a trajectory, wherein the vehicle (1) has a control device (2) which plans the trajectory within a definable search space of the trajectory and has access to actuators (3, 4, 5) of the vehicle (1) in order to control the vehicle (1). At least one limit value is determined for at least one control variable of the actuators (3, 4, 5), and the search space (9) of the control variable is limited using the at least one limit value, wherein the trajectory is planned using the search space (9).

Description

Technical Field

[0001] This invention relates to a method for controlling a vehicle along a trajectory. It also relates to a control device designed to control a vehicle along a trajectory using the method according to the invention, a computer program having program code for executing the method according to the invention, and a computer-readable storage medium that causes a computer to execute the method according to the invention, the computer-readable storage medium being executed on the computer. Background Technology

[0002] Modern vehicles, such as passenger cars, trucks, motorized two-wheelers, or other modes of transport known from the prior art, are increasingly equipped with driver assistance systems. These systems, with the help of appropriate sensors or sensor systems, can detect the environment, identify traffic conditions, and assist the driver, for example, by intervening with braking or steering, or by issuing visual or auditory warnings. Radar sensors, lidar sensors, camera sensors, ultrasonic sensors, and the like are frequently used as sensor systems for detecting the environment. Based on the sensor data determined by the sensors, conclusions about the environment can be drawn. These conclusions can then be used to implement general assistance functions, such as lane keeping control or lane keeping assist (LKA).

[0003] Furthermore, modern vehicles typically include electric power steering or steering assist (EPS = Electric Power Steering, EPAS = Electric Power Assisted Steering) or assists driver power steering by reducing the force required from the driver to operate the steering wheel. This can be achieved, for example, by using an electric servo motor (EPS motor or electric power steering motor) or servo motor mounted on a mechanism in the steering system (e.g., on the steering column or steering transmission), and by assisting or superimposing the driver's steering movements with the applied motor torque or servo torque. In this case, the electric servo motor and associated control unit can be positioned within the steering system (C-EPS or column-assisted EPS), on the pinion gear of the steering transmission (P-EPS or pinion-assisted EPS), or parallel / concentrically around the rack (R-EPS or rack-assisted EPS). In addition, a sensor system is provided, which includes an absolute steering wheel angle sensor, a steering torque sensor, and a relative rotor position angle sensor of the motor, as well as (if necessary) a current sensor, which can be used to estimate motor torque or servo torque, for example.

[0004] In the fields of driver assistance and autonomous driving, vehicles are typically controlled via a series of planners (e.g., motion planners and trajectory planners) and controllers. In this scenario, the controller attempts to follow a trajectory generated by the planners. However, depending on the planning method, it is possible that the trajectory planned by the planner cannot, or can only be very difficult to, be driven by the controller or the vehicle due to actuator limitations that the planner is unaware of or unable to handle. The actuators may therefore reach their limits, potentially leading to a wind-up effect in the control. This means that during the planning cycle, the vehicle may deviate from the desired trajectory, requiring the newly planned trajectory to be modified further to compensate for this deviation. This creates internal dynamics between the planner and controller, which degrades the vehicle's managed behavior and can be noticed by the driver, such as swaying within the lane. If the trajectory cannot be driven, the driver must intervene. Model prediction methods exist for planning and / or control, which have the advantage of considering vehicle dynamics and actuator limitations in the form of a model during planning. Using this method, if the model is accurate, lower-level controllers can be omitted. However, as model complexity increases, such methods require more computational effort, which is already very high for simple models. As a result, it is impractical to consider the dynamics and constraints of the actuator in addition.

[0005] DE 10 2016 221 723 A1 discloses a control system for a vehicle having multiple actuators or actuators (e.g., steering system, powertrain, service brake, and parking brake). In this case, the control system includes modules for controlling vehicle motion, modules for controlling the actuators, modules for specifying a vehicle operating strategy to be implemented, and modules for coordinating torque, wherein motion requirements applied to the vehicle are used to form a normalized requirement vector that results in longitudinal, lateral, and vertical components. Furthermore, the control system is configured to form torque distributed among the actuators based on the vehicle operating strategy and the requirement vector.

[0006] DE 10 2015 209 066 A1 describes a method for cost reduction in trajectory planning for vehicles, in which the search space for determining the trajectory is limited based on an approximate end time. In this case, the search space for determining the trajectory for driving maneuvers is limited to a specific range before and after the approximate end time, specifically within 10% of the approximate end time, in order to reduce the computational effort required to determine the trajectory. Summary of the Invention

[0007] Therefore, the present invention aims to provide a general method for controlling vehicles, in which trajectory planning is improved in a simple and cost-effective manner.

[0008] In the method for controlling a vehicle along a trajectory according to the invention, the vehicle has a control device that plans a trajectory within a definable search space (the search space of the trajectory or the search space for trajectory planning) and has access to the vehicle's actuators to control the vehicle, wherein at least one limit value is determined for at least one control variable of the actuator, and the search space of the control variable is defined based on one or more limit values. The trajectory is then planned using the search space of the control variable. In this case, the search space of the control variable constitutes a subspace of the search space of the trajectory, meaning that limiting the search space of the control variable also implicitly limits the search space of the trajectory. The method according to the invention can therefore calculate the trajectory dynamically related to the actuator. Furthermore, this can save significant computation time, for example, compared to methods that dynamically integrate the actuator into an MPC (model predictive control) model. Due to the fact that this method is separate from the planner, it can also be used with different planner methods, thus easily taking into account software limitations of motor torque and steering speed, as well as deterioration of the steering system. This further improves operational reliability. In addition, the described method can also be used with multiple actuators and is independent of complex vehicle models.

[0009] Preferably, the EPS motor of the vehicle's steering system or electric steering system is used as the actuator.

[0010] Steering angle and / or steering angular velocity and / or road curvature and / or EPS motor torque can be easily set as control variables.

[0011] Advantageously, the maximum time progression to the left and the maximum time progression to the right of the control variables can be set to limit values, and then these maximum time progressions can be coordinated with the planner. If the steering angle is set as a control variable or is one of the control variables, this can preferably be the maximum steering angle to the left and the maximum steering angle to the right. Alternatively or additionally, the maximum drivable curvature (of the lane) to the left and the maximum drivable curvature (of the lane) to the right can also be set to limit values.

[0012] According to a further advantageous configuration of the invention, the difference between the force currently applied to the EPS motor and the maximum available force can be determined, for example, by the fact that the power of the EPS or the EPS motor is initially set or can be predicted by the EPS as an input signal. This has the advantage that different levels of degradation can also be represented, for example, if only a portion of the power is available. Therefore, for example, the potential of the EPS motor and / or the power of the EPS motor can be estimated by using the increasingly larger difference to indicate an increasingly larger level of degradation.

[0013] Furthermore, the nonlinear friction force of the steering system can be determined, wherein the limit value is determined by taking into account the nonlinear friction force.

[0014] Road forces can be easily estimated, and the result is that the limit values ​​can be determined while taking road forces into account.

[0015] In this case, road forces can be determined using virtual spring-based modeling.

[0016] The spring stiffness of a virtual spring can be advantageously determined or calculated using vehicle speed and motor torque. For example, the spring stiffness can be described using a mathematical term consisting of a first part that is purely speed-related (e.g., from vehicle speed) and a second part that is speed-related and torque-related (e.g., from vehicle speed and maximum available motor torque or EPS torque).

[0017] In this case, an estimation method, such as least squares, and particularly recursive least squares (RLS), is preferably used to determine the spring stiffness. However, alternatively, other estimation methods can also be used. For example, an initial estimate can be provided, which can be performed offline and does not require a recursive method (such as RLS). Although this approach is independent of the method flow, it may heavily depend on other parameters, such as the tires used.

[0018] In addition, at least one sensor can be set up for detecting the surrounding environment, particularly a camera and / or lidar sensor and / or radar sensor and / or ultrasonic sensor. Sensor data from one or more sensors can be used to detect the vehicle environment and objects and road users located within it. In this case, sensor data from multiple sensors can also be fused to improve the detection of the environment and objects.

[0019] In a practical manner, the detected environment surrounding the vehicle (including objects and road users within it) can be used to define the control variables and / or the search space for trajectory planning. This can be done, for example, by further restricting the search space of possible trajectories, since the objects detected by the sensors are located within the previously restricted search space. Furthermore, the trajectory to be followed can be selected during or after trajectory planning in a manner that takes into account collision avoidance, for example, by selecting a route that extends along the road and does not collide with other objects / road users.

[0020] The present invention also includes a computer program having program code that, when executed in a computer or in another programmable computer known in the prior art, performs the method according to the invention. Therefore, the method can also be in the form of a purely computer-implemented method, wherein, in the sense of the present invention, the term "computer-implemented method" describes a method flow or process implemented or performed using a computer. A computer, such as a computer network or other programmable device known in the prior art (e.g., a computer device including a processor, microcontroller, etc.), can in this case process data using programmable computational rules.

[0021] Furthermore, the present invention also relates to a computer-readable storage medium including instructions that cause a computer to perform the methods described above, and these instructions are executed on the computer.

[0022] Parallel or subordinate, the present invention also includes a control device for controlling a vehicle along a trajectory, the control device being configured to control the vehicle using the method according to the invention.

[0023] Within the meaning of this invention, the term "search space for vehicle trajectory or for trajectory planning" is understood as the spatial and temporal extent in which the control unit searches for possible drivable trajectories, wherein multiple trajectories can be planned within the search space so as to select the appropriate trajectory for the situation. Within the meaning of this invention, the term "search space for control variables" is understood as the spatial and temporal extent in which the control unit searches for possible control variables. In this case, the search space for control variables constitutes a subspace of the search space for vehicle trajectory.

[0024] Within the meaning of this invention, the term "limit value" is understood as the maximum or minimum value of a control variable, i.e., the maximum or minimum value at which its progress, for example, along the distance traveled or time t, can be detected.

[0025] This invention also explicitly includes features or combinations of features not explicitly mentioned, and so-called sub-combinations. Attached Figure Description

[0026] The invention will now be described in more detail with reference to advantageous exemplary embodiments. In this case:

[0027] Figure 1 A simplified schematic diagram of a vehicle is shown, in which the maximum control variable is predicted using a method according to the invention;

[0028] Figure 2 A simplified schematic diagram illustrating the correlation between virtual stiffness and vehicle speed and maximum EPS torque is shown.

[0029] Figure 3 A simplified diagram of the search space for the steering angle limited by the method according to the invention is shown, and

[0030] Figure 4 A simplified schematic diagram of the flowchart of the method according to the present invention is shown. Detailed Implementation

[0031] Figure 1 Reference numeral 1 in the attached figure indicates a vehicle with various actuators (steering system 3, power unit 4, brakes 5) and a control device 2 (ECU, electronic control unit) that can dynamically perform trajectory planning for one or more actuators. In this case, the trajectory is calculated using a trajectory planner, where the maximum control variable for the corresponding actuator is predicted, particularly in the lateral direction relative to the search space constraints of the trajectory planner, and this prediction of the maximum control variable is used for trajectory planning. In this case, the trajectory planner can be in the form of a hardware module of the control device 2 or a purely software module. The vehicle 1 also has sensors for detecting the surrounding environment (camera 6, lidar sensor 7, and radar sensor 8). Sensor data from these sensors is used to identify the surrounding environment and objects, resulting in the implementation of various assistance functions, such as emergency braking assist (EBA, electronic brake assist), adaptive cruise control (ACC, automatic cruise control), lane keeping control or lane keeping assist (LKA), etc. In practice, the assistance functions can also be performed via the control device 2 or a dedicated control device.

[0032] The method according to the invention is generally applicable to all actuators found in general-purpose vehicles, and therefore also to all types of steering systems used in general-purpose transportation vehicles. Thus, the method can also be applied to vehicles with additional actuators, i.e., vehicles that also have front and rear axle steering. The method according to the invention is described below with reference to a vehicle with front axle steering, wherein the steering angle δ is used as a control variable, i.e., the current steering angle δ can be measured first as an initial value. It can be assumed here that the trajectory planner method used can handle the maximum steering angle δ_max. However, alternatively or additionally, other control variables, such as steering angular velocity or curvature, may also be used. The force balance in the steering system can be described mathematically as follows:

[0033] (1) m_EPS·a=F_Mot-d·v-F_friction-F_load, where m_EPS is the cumulative mass of the steering system, a is the acceleration of the rack, F_Mot is the force provided by the EPS motor, d is the damping of the EPS, v is the speed of the rack, F_friction is the nonlinear friction of the EPS, and F_load is the load applied to the EPS, which includes the road force F_Str and the force from the steering wheel. For example, when driving on a road, such road forces are applied to the vehicle wheels. To dissipate the energy transfer of these road forces, spring or damper assemblies are typically used in the vehicle suspension system.

[0034] The maximum control variable is conveniently determined based on the actuator power available in the absence of any disturbances. In this case, disturbance variables, such as the driver's manual torque included in the force from the steering wheel, can be ignored. External disturbance variables, such as crosswinds, are ignored because such disturbances can be compensated for, for example, by control. On the other hand, the remaining road forces F_Str and therefore F_load cannot be easily ignored because they have a significant impact on the maximum steering angle and are therefore not considered disturbances since they always occur. Observing the road forces F_Str at the vehicle level in the single-track model shows that these road forces depend on the current steering angle, vehicle speed, and road friction coefficient. However, the effect of the road friction coefficient can be ignored here, resulting in only considering scenarios with a high friction coefficient. This is achievable because while a reduced friction coefficient would result in a higher maximum steering angle, it does not necessarily result in a higher drivable curvature and therefore no drivable trajectory. Accordingly, the dependence of the road forces F_Str on the steering angle and vehicle speed still exists. Due to the steering angle dependence used to model the road forces F_Str, a virtual spring with a spring stiffness c related to the vehicle speed is used. The spring stiffness c also depends on the maximum set EPS torque M_Mot_max, such as Figure 2As shown. This is caused by nonlinearity (such as the gear ratio between rack travel and wheel steering angle, or subsequent operation related to steering angle).

[0035] Therefore, the following result is obtained from equation (1):

[0036] (2) m_EPS·a=F_Mot-d·v-F_friction-c(v_veh,M_Mot_max)·x.

[0037] In this case, v_veh is the vehicle speed, and x is the rack position, which can be converted into a steering angle δ using the gear ratio i. In this case, the term c(v_veh, M_Mot_max) consists of a purely speed-related part c1(v-veh) and a speed-related and torque-related part c2(v_veh, M_Mot_max):

[0038] (3) c (v_veh, M_Mot_max) = c1 (v_veh) + c2 (v_veh, M_Mot_max).

[0039] This can be used to derive a lookup table for the spring stiffness c (based on...). Figure 2 For example, this lookup table can be derived from the step excitation of the steering system at different speeds. For instance, the spring stiffness c can be estimated and adjusted based on speed using the RLS (Recursive Least Squares) algorithm. In this case, only c1(v_veh) needs to be adjusted, because the term c2(v_veh, M_Mot_max) reflects a structural, invariant relationship. Therefore, in this case, the maximum settable steering angle δ to the left and right should be predicted. In this case, the maximum force that the EPS motor can still set for F_Mot is selected and applied as a step, which is filtered by the motor time constant T_Mot according to the following:

[0040] (4)F_Mot=1 / (T_Mot·s+1)·F_Mot_max.

[0041] In this case, F_Mot_max is the difference between the currently applied force and the maximum available force. The maximum available force can be determined using the power of the EPS, or it can be predicted using the EPS as an input signal. For example, if only a portion of the power is available, this can represent different levels of degradation of the EPS motor. Nonlinear friction F_friction corresponds to the static friction in the system and can also be considered via the so-called dead zone in the motor force F_Mot, since only a constant direction of motion is considered, and the hysteresis effect of static friction therefore does not take effect. Thus, the following equation is derived:

[0042] (5)F_Mot_Fric = 0 if |F_Mot| < F_Adhesion

[0043] F_Mot - F_Adhesion if F_Mot > F_Adhesion

[0044] F_Mot + F_Adhesion if -F_Mot < F_Adhesion.

[0045] Here, F_Adhesion is the magnitude of the adhesion force. Additionally, Equation (6) is obtained from Equation (2):

[0046] (6)m_EPS·a = F_Mot_Fric - d·v - c(v_veh, M_Mot_max)·x, which corresponds to a second-order delay link. In this case, the damping d can be selected as a constant.

[0047] Therefore, the maximum rack position or the maximum steering angle δ_max can be predicted by performing a transformation and double integration based on the acceleration in Equation (6). Then, the obtained vectors of the maximum steering angle to the right (δ_max_re) and to the left (δ_max_li) over time t can be sampled in order to reduce the amount of data to be sent and forwarded to the planner as a feedback signal. These two vectors indicate the upper and lower limits of the search space of the control variable, within which the trajectory planner can search for an optimal solution, as Figure 3 shown, using the finite search space 9, which is shown by the dashed line between the two vectors δ_max_re, δ_max_li.

[0048] In an exemplary embodiment of the method flow according to Figure 4 , the maximum time progress of the predicted steering angles to the left and to the right for a vehicle with front axle steering is output. In this case, the steering angle δ is first determined or measured as a starting value (steering angle determination 12), and for example, the described look-up table is used (see Figure 2The spring stiffness c is determined (determination of spring stiffness 10). Furthermore, motor characteristics and features 11a (left side) and 11b (right side) are determined, particularly based on the currently applied motor torque (motor torque detection), i.e., the currently applied motor torque M_Mot. Motor characteristics and features may differ to the left and right, for example due to steering system asymmetry or artificially introduced asymmetry, such as during LDP (Lane Departure Protection) functionality, where steering is more restricted in the direction closer to the lane boundary. The progression of the maximum control variable can then be predicted based on the spring stiffness and the motor characteristics and features, in this case, the maximum steering angle to the left (prediction for left side 13) and the maximum steering angle to the right (prediction for right side 14). The predicted steering angles are then forwarded to the planner 15. If the vehicle also has rear axle steering, the rear axle steering angle can be determined in the same manner as the front axle steering angle, i.e., there are two additional vectors on the rear axle for rear axle steering, one for the maximum steering angle to the left and one for the maximum steering angle to the right. Therefore, Figure 4 The method flow described in the document outlines the process for determining the maximum steering angle of the front or rear axle. Alternatively or additionally, the curvature to be driven can be used as a control variable, regardless of whether rear axle steering is present. The advantage of this configuration is that only two vectors (maximum curvature to the left and maximum curvature to the right) are generated, even if rear axle steering is present. However, a vehicle model should then be provided again for determination.

[0049] In a practical way, the predicted limits of the control variables can also be used for the concept of anti-saturation in controllers. Furthermore, it is also possible to use... Figure 2 The load on the vehicle is estimated using a lookup table of relationships and stiffness.

[0050] List of reference numerals in the attached diagram:

[0051] 1 vehicle

[0052] 2. Control equipment

[0053] 3. Steering System

[0054] 4. Power equipment

[0055] 5. Brakes

[0056] 6 cameras

[0057] 7. LiDAR Sensor

[0058] 8. Radar Sensors

[0059] 9. Search space (for control variables or steering angles)

[0060] 10. Determining the stiffness of a spring

[0061] 11a Motor characteristics and features (left side)

[0062] 11b Motor characteristics and features (right side)

[0063] 12. Determining the steering angle

[0064] 13. Prediction of the maximum steering angle on the left side

[0065] 14. Prediction of the maximum steering angle on the right side

[0066] 15. Trajectory Planner

[0067] δ steering angle

[0068] c Spring stiffness

Claims

1. A method for controlling a vehicle (1) along a trajectory, wherein, The vehicle (1) has a control device (2) that plans the trajectory within a defined search space and has access to the actuators (3, 4, 5) of the vehicle (1) to control the vehicle (1), wherein, Determine at least one limit value for at least one control variable of actuators (3, 4, 5). The search space (9) of the control variable is defined based on at least one limit value, wherein, The trajectory is planned using the search space (9), wherein the steering angle and / or steering angular velocity and / or road curvature and / or EPS motor torque are set as the control variables, the maximum time progression to the left and the maximum time progression to the right of the control variables are set as the limit values, and these maximum time progressions are coordinated with the planner, and the EPS motor of the steering system (3) is set as the actuator, characterized in that the difference between the force currently applied to the EPS motor and the maximum available force is determined, and the difference is used to estimate the potential of the EPS motor.

2. The method according to claim 1, characterized in that, Determine the nonlinear friction force (F_Fric), and determine the limit value while taking into account these nonlinear friction forces (F_Fric).

3. The method according to claim 1 or 2, characterized in that, Determine the road forces (F_Str), and determine the limit value while taking these road forces (F_Str) into account.

4. The method according to claim 3, characterized in that, These road forces (F_Str) are determined using virtual spring-based modeling.

5. The method according to claim 4, characterized in that, The spring stiffness (c) of the virtual spring is determined by the vehicle speed and motor torque.

6. The method according to claim 5, characterized in that, The spring stiffness (c) is determined using the least squares method.

7. The method according to claim 1 or 2, characterized in that, Install at least one sensor for detecting the surrounding environment.

8. The method according to claim 7, characterized in that, Use the detected environment to define the search space (9) and / or for trajectory planning.

9. The method according to claim 6, characterized in that, The spring stiffness (c) is determined using the recursive least squares (RLS) method.

10. The method according to claim 7, characterized in that, At least one sensor is a camera (6) and / or a lidar sensor (7) and / or a radar sensor (8) and / or an ultrasonic sensor.

11. A computer program product having program code, wherein when the program code of the computer program product is executed on a computer, the program code performs the method according to any one of the preceding claims.

12. A computer-readable storage medium comprising instructions that cause a computer to perform the method according to any one of claims 1 to 10, wherein the instructions are executed on the computer.

13. A control device (2) for controlling a vehicle (1) along a trajectory, characterized in that, The vehicle (1) is controlled by the method according to any one of claims 1 to 10.

Images